Jump to content

Render times and patch cable lengths...


Recommended Posts

I've just finished setting up 2 new systems, both have exactly the same spec, and run max 2008 with vray. I was in the process of benchmarking the machines against the older machines in the office by rendering out the same scene, and decided to check that the new systems were pumping out identical (or near enough) render times. I was suprised to find that one machine was 17 seconds faster than the other?

 

As I had set up both machines myself, I was sure that nothing I had done was different in any way, but my attention was drawn to the 10m cat 5e patch cable which connected one new machine to the 8-port switch. The cable connecting the other new machine to the switch was only 1.5m, I asked my mate and we both agreed that surely the 8.5 meter difference could not be accountable for the difference in speed?

 

To be sure I swapped the cables, and sure enough, the computer which had the longer cable suffered the same symptoms, a 17 second delay in render time, now this delay was once I'd hit render, at the loading bitmaps stage, so most of the delay was the pause when looking for bitmaps, this didn't happen on the machine with the sharter cable.

 

I checked that both cables were Cat 5e patch and that the cables weren't faulty. Is this symptom normal with long cables?

 

Thanks

Edited by Brute Guy
Link to comment
Share on other sites

Hmm... that's interesting.

 

Well, one cable is 666% times longer than the other.

Back in the days of NetWare (shudder), the last computer on the segment was a bit further than the rest, and it never worked right.

 

Attentuation? 666% more resistance on the network line?

 

Are these cables both out of the original packaging? New??

I'd spend another $5 and get new cables from a computer shop and try again.

 

IF the cable is making a difference, then it would only be if it was a net render like AJ suggests.

 

If you ping both machines from the server, the one with the longer network cable should have a higher ping time.

Link to comment
Share on other sites

Hi SandmanNinja,

 

One is fairly old, I think what your saying is a good start, I have 3 new on order coming soon, so i'll check the difference when they arrive. Come to think of it, I did do a distributed render test earlier using the 2 machines, however the render paused... then crashed after the first prepass, is this what you both are suggesting when you mention net render being a possible problem?

 

Marc

Link to comment
Share on other sites

If there's a problem with the cable, it could cause a drop in signal/noise ratio resulting in lost packets which would cause delays as your computer and the server resend data to correct the problem. This isn't an issue in casual use but in net rendering, transferring a lot of maps and other intensive operations it could be a problem. This is why the Cat5 spec is really tight as far as things like preserving the twisting of the wires.

 

It's possible to test for packet loss using software that sends a bunch of pings to a server but I don't have a link to one...

Link to comment
Share on other sites

Oh -- isn't it obvious.

 

With one cable being longer than another, the longer cable creates a larger mangnetic field; especially, when you consider that the longer cable is coiled up. The resultant magnetic field is working to realign the earths magnetic field. That change in the magnetic field is increasing the potection ratio of the ionosphere and ozone layer. The resultant reduction in solar radiation gain is decreasing the efficiency of the chips in your system due to the differential in heat gain and ongoing electromagnetic resistance.

 

Simple really....

Link to comment
Share on other sites

I think Ted's got it. When I first read the post, I had neglected to take into account that the cable is 10 meters and as the meter is based on the speed of light and the UV radiation Ted suggests also travels at the speed of light, any power of 10 m will set up a harmonic resonance loop that will produce the effect described. (The UV radiation is solar; this is why the problem only occurs in the day time.)

 

This is one more very good reason why we should use imperial units for all calculations and equipment! It is very easy to forget that Metric originated as a Communist plot, now that communism as a global menace is out of style, but the nefarious side effects linger.

Link to comment
Share on other sites

Perhaps one of the computers was on a desk and the other was on the floor. I know I go a *lot* slower uphill... why wouldn't the same be true for electrons in a patch cable? Maybe there was a kink in one of the cables... you know, the garden hose effect.

 

I think a couple of my brain cells just walked out in protest.

Link to comment
Share on other sites

Yeah, Cat6 is rated for a gigabit/sec, and the quality is much higher. I use them for all mission-critical apps (i.e. my PC with Crysis on it). I (think) it's got some special criterial to be qualified to be called Cat-6, same manufacturer of the connectors, etc...

Link to comment
Share on other sites

Yeah, Cat6 is rated for a gigabit/sec, and the quality is much higher. I use them for all mission-critical apps (i.e. my PC with Crysis on it). I (think) it's got some special criterial to be qualified to be called Cat-6, same manufacturer of the connectors, etc...

 

Brute - Any update? Is your switch Giga-speed? At these short lengths, even Cat 5e will pass gigaspeed?

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share

×
×
  • Create New...